asucicrepo / patagonia-ocr

ASU-CIC OCR API

Description

The API receives a single file in the file key of a multipart/form-data POST payload, validates it, converts it if necessary, and uploads it to S3.

When the provided file is a PDF document a call is made to AWS Textract and the HTTP request is responded immediately with a requestId, the process continues asynchronously via an SNS topic, when the AWS Textract job is complete, the obtained data is mapped, validated, normalized, and then saved in a .csv file in an S3 bucket. Later a GET request can be made to the same endpoint appending the /requestId to the url, to retrieve the result data when the process is finished.

On the other side, when the provided file is an image the process is carried on synchronously, and the result data will be available in the same POST response.

Currently Supported file types

File posted can be any PNG / JPG / HEIC image or a PDF document. Maximum file size is 6MB.

Currently Supported documentTypes

Utility Bills

• APS
• Southwest Gas
• City of Phoenix Water
• SRP

Others

• Arizona Driver license

Important Notes

1- Objects stored in the S3 bucket have a lifecycle of 14 days before deletion.

2- You can add ?debug to any POST or GET request made to view more details on output and logs. Also all intermediate states of the data is saved into the S3 bucket when this flag is present.

---

API Documentation

See README.api.md for full docs regarding the API.

Data Extraction Logic Documentation

See README.ext.md for docs on currently supported documentTypes

Development Documentation

See README.dev.md